[c-nsp] arp timeout issues

[Rodney Dunn]

But you don't see it added back?

Change the timestamp method on the event trace to msec granularity
and capture one for me.

I think it's ip cef event CLI parse chain to change the timestamp.

We should not delete the arp entry on a successful refresh.

What I really need is a debug arp at the same time you get
a refresh and the corresponding CEF event log there.

Rodney

On Mon, Feb 26, 2007 at 09:51:57AM -0000, Ed Butler wrote:
> Yes, when the ARP times out we see this:
> 
> sh ip cef event | in 1.2.3.4
> 1y5w:  [Default] 1.2.3.4/32'08 FIB remove (flagged)     [OK]
> 1y5w:  [Default] 1.2.3.4/32'08 FIB remove (deleted)     [OK] 
> 
> 
> Regards,
> 
> Ed Butler
> RapidSwitch Ltd
> DDI: 020 7106 0731
> 
> RapidSwitch Ltd, Sovereign House, 227 Marsh Wall, London, E14 9SD
> 
> This email message is intended only for the addressee(s) and contains
> information that may be confidential and/or copyright.  No
> representation is made that this email or any attachments are free of
> viruses. Virus scanning is recommended and is the responsibility of the
> recipient. RapidSwitch Ltd is registered in England & Wales Company No.
> 04535693, Registered Address: 3 Beacon Mews, South Road, Weybridge, KT13
> 9DZ. VAT Registration No. GB 794 3811 96. All prices exclude VAT unless
> otherwise stated. Quotations valid for seven days.
> -----Original Message-----
> From: Rodney Dunn [mailto:rodunn at cisco.com] 
> Sent: 25 February 2007 18:16
> To: Ed Butler
> Cc: Cisco-NSP Mailing List
> Subject: Re: [c-nsp] arp timeout issues
> 
> I don't know about those switches but can you see if there is a 'sh ip
> cef event' and see if the adj is being updated when the arp refresh
> happens?
> 
> I wonder if the hw adj is being updated and while we lock the adjacency
> to update it we don't switch packets to it.
> 
> Have you put a sniffer on there to make sure the packets are not going
> out?
> 
> 
> On Sat, Feb 24, 2007 at 08:07:30PM -0000, Ed Butler wrote:
> > Fairly standard set up; stack of 3750s running OSPF to 6500s. On the 
> > 3750s there are a number of SVIs, with "arp timeout 300" set on them.
> > The same VLANs have a MAC againg time of 5 minutes too.
> > 
> > The behaviour we are seeing is that every six minutes (not 5) traffic 
> > to a host on the SVI will die for 500ms or so, and this coincides with
> 
> > the ARP entry being refreshed. CPU usage on the 3750 stack averages
> 15-20%.
> > 
> > We have a number of similar set up that don't replicate this problem.
> > Has anyone seen similar and can recommend a course of action? 
> > Hopefully something more sophisiticated than simply increasing the ARP
> timeout!
> > 
> > Regards,
> > 
> > Ed Butler
> > RapidSwitch Ltd
> > DDI: 020 7106 0731
> > 
> > RapidSwitch Ltd, Sovereign House, 227 Marsh Wall, London, E14 9SD
> > 
> > This email message is intended only for the addressee(s) and contains 
> > information that may be confidential and/or copyright.  No 
> > representation is made that this email or any attachments are free of 
> > viruses. Virus scanning is recommended and is the responsibility of 
> > the recipient. RapidSwitch Ltd is registered in England & Wales
> Company No.
> > 04535693, Registered Address: 3 Beacon Mews, South Road, Weybridge, 
> > KT13 9DZ. VAT Registration No. GB 794 3811 96. All prices exclude VAT 
> > unless otherwise stated. Quotations valid for seven days.
> > 
> > _______________________________________________
> > cisco-nsp mailing list  cisco-nsp at puck.nether.net 
> > https://puck.nether.net/mailman/listinfo/cisco-nsp
> > archive at http://puck.nether.net/pipermail/cisco-nsp/
> 
> 
> _______________________________________________
> cisco-nsp mailing list  cisco-nsp at puck.nether.net
> https://puck.nether.net/mailman/listinfo/cisco-nsp
> archive at http://puck.nether.net/pipermail/cisco-nsp/